Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: bootstrap weights


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: bootstrap weights
Date   Fri, 25 Nov 2011 18:55:53 -0600

Cam,

that's an overkill. You are trying to outsmart the data by being to
conservative.

Stefano,

that's probably the best you can do. I would suggest using more
replications (reps() option) and fewer subsampled units (I tend to use
n(-1), although with several thousand observations that you probably
have, the small sample performance these adjustments tend to improve
does not matter). Ask your data providers if they can generate the
bootstrap weights on their side using their strata information.

On Fri, Nov 25, 2011 at 3:32 PM, Cameron McIntosh <cnm100@hotmail.com> wrote:
> Stefano,
>
> I wonder if another, conservative approach might be to form clusters that maximize ICC, or more directly the intra-cluster covariances of the parameter estimates, and then use Taylor linearization to get your adjusted standard errors. It may take a bit of work to write an algorithm to generate clusters such as I describe.
>
> For what it's worth,
>
> Cam
>
>> Date: Fri, 25 Nov 2011 19:39:43 +0000
>> Subject: Re: st: bootstrap weights
>> From: verdes@tcd.ie
>> To: statalist@hsphsun2.harvard.edu
>>
>> Dear Bryan and Steven,
>>
>> thank you very much for your help.
>>
>> Only today the national statistics office confirmed to me that: 1) no
>> clusters were used in the household survey I have, and 2) they cannot
>> release information about strata.
>> Thus, my microdata set only includes sampling weights (which sum up to
>> Ireland's household population, N).
>>
>> Given this, what is the best I can do? I have thought of the following:
>>
>> svyset _n [pw = my_sampling_weights]
>> bsweights bsw, n(0) reps(100)
>> bs4rw exp_list, rw(bsw*): command [pw = my_sampling_weights]
>>
>> The first line - on Steven's suggestion - states that there is a
>> single stratum only and each household is a PSU.
>> The second line creates bootstrap resampling weights, which are then
>> used in the bootstrap of exp_list (-bsweights- and -bs4rw- are
>> commands written by Stas Kolenikov).
>>
>> I'm aware that not having information on strata affects results.
>> However, my two questions are:
>>
>> 1) Do sampling weights mitigate this error?
>>
>> 2) Are the three command lines above the best I can do, given the
>> information limitations?
>>
>> Once again I'd greatly appreciate your help.
>>
>> Best regards,
>> Stefano Verde
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>



-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index